This paper reports results from three prospective studies, in which probands were followed over periods of 10 years, before inquiring about death and cause of death. Personality inventories were administered at the beginning of the 10 year period, as were questions concerning smoking, drinking, medical diseases, etc. It was found that personality variables were much more predictive of death from cancer or cardiovascular disease than was smoking, and that different personality types were susceptible to either of these two diseases. Personality type was defined in terms of differential ways of dealing with interpersonal stress, and it was found that stress was a very potent cause of death, in the sense that stressed probands had a 40 per cent higher death rate than non-stressed probands.